MCMC-based inversion algorithm dedicated to NEMS mass Spectrometry

Nano Electro Mechanical Systems (NEMS) provide new perspectives in the mass spectrometry field. This new generation of sensors is sensitive enough to detect a single molecule. Thus, it is possible to estimate a concentration profile in a counting-mode which brings a reduced noise and a higher sensitivity. In this paper, first, we briefly describe the measurement system. Then we propose a probabilistic model of the acquisition system in the form of an input-output system from which we can deduce the likelihood of the unknowns in the data and a Bayesian inference approach with a hierarchical Bernoulli-Gamma prior model. To do the computation we propose the use of a Multiple-Try Metropolis Monte-Carlo Markov-Chain algorithm. Multiple-Try Metropolis proposal functions are adapted to the model, especially to the discrete nature of the problem. Our approach provides an automatic robust estimation of mass spectra. We test the proposed algorithm both on experimental and on simulated data. We discuss the performan...